Convex unknown input observer for sensor fault estimation of nonlinear descriptor systems via linear matrix inequalities

This paper proposes a novel convex unknown input observer for simultaneously estimating system states and sensor faults in nonlinear systems. The observer design is based on the Lyapunov method and provides conditions in the form of linear matrix inequalities. The methodology allows handling a larger family of systems than previous approaches as it considers unknown premise variables for the convex error system. Numerical simulations and real‐time experiments demonstrate the effectiveness in practical applications. The results suggest that the proposed convex observer can accurately estimate system states and sensor faults.